Spiekermann, Sarah: Online Information Search with Electronic Agents: Drivers, Impediments, and Privacy Issues


Chapter 5. Comparing Online Search Behavior for Different Product Categories

In order to investigate the extend to which perceived purchase risk would influence the use of agents or motivate manual search, shopping behavior was compared for two different product groups for which it was expected to measure different levels of purchase risk: compact cameras and winter jackets. The belief that compact cameras and winter jackets would be perceived differently by experimental participants was based on them being search and experience goods (see below).

In the 1990s a distinction of search, experience, and trust goods developed in information economics [Darby and Karni, 1973; Nelson, 1970] has found an entry into marketing literature of institutional theory [Arnthorsson, 1991; Kaas, 1990, 1995; Weiber and Adler, 1995a, b]. Products with strong search characteristics are distinguished by the fact that they can be fully judged by inspection or equivalent information search prior to purchase. Products with dominant experience characteristics can only be fully judged after purchase and use. They are thus implying a higher purchase risk than do search goods, because the buyer‘s expectations might be disappointed [Weiber and Adler, 1995b]. Finally, products with trust characteristics are marked by the fact that their quality can neither be judged on before nor after the purchase. Given the proclaimed relationship between product nature and risk, compact cameras and winter jackets were chosen for the current experiment, assuming that they could be considered as relatively good representatives for search and experience goods. Compact cameras usually entail strong search good characteristics as their quality can be well described prior to purchase on the basis of product attributes. In contrast, jackets were considered to be a typical experience good, because one has to wear them and feel the model before assessing the fit.


Based on the observations made in the structural equation model presented above, it was hypothesised that subjects shopping for winter jackets would rely more heavily on manually controlled search than camera shoppers.

5.1 Empirical Survey Design

5.1.1 Data

All treatments summarized in table 1 were originally included in the navigational analysis. However, 11 of the 206 participants had missing data relevant for the analysis and one individual answered the wrong questionnaire. This led to a dataset of 195 observations.

In order to investigate the relatively isolated effect of product nature on interaction it was necessary to respect individual factors that could potentially have had a strong influence on interaction, but are independent from product. The dataset of 195 observations (144 cameras, 51 jackets) was therefore investigated and straightened out with a view to three factors that the structural equation model had revealed to influence interaction apart from product: namely privacy concerns, satisfaction with the agent‘s recommendations and perceived time cost.

An in-depths analysis of privacy concerns revealed that most subjects, even though they stated to be privacy conscious, did not act accordingly [Spiekermann et al., 2001]. Yet, for the purpose of the current research it is important to note that only 3 participants (2 camera shoppers, 1 jacket shopper) expressed considerable privacy concern before entering the online store and also acted consistently with their expressed attitude by refusing most of the interaction with the shopbot (see table 14). These subjects have been excluded from the current analysis. Their behavior cannot be interpreted as a response to the product.

Furthermore, the perception of the search engine‘s accuracy had a significant influence on interaction. It was measured by asking participants after the shopping session how valuable and accurate they had found the agent‘s product recommendations. While 78,1% of the participants (group 1) felt that the search


engine made either accurate (7,8%), quite accurate (29,2%) or at least accurate (41,1%) recommendations, 21,9% were not fond of the search aid (group 2). Mann-Whitney-U-Test used to investigate the impact of this distinct search engine perception on the total number of page requests yielded significant differences for the two perception groups (z = - 2.716, p = 0,007). As a consequence, 23 camera shoppers and 19 jacket shoppers have been excluded from the analysis presented hereafter (for SPSS output file see Appendix D, table D3).

Finally, following the shopping session participants were asked whether they had rather done something else instead of shopping for a compact camera or winter jacket in our experimental store (measured as time cost in the structural equation model). 6 subjects admitted a relatively strong de-motivation.<31> Mann-Whitney-U-Test for the impact of this de-motivation on the total number of page requests, however, did not yield significant differences in behavior (z = -.341, p = .733), nor did a T-test on the time spent shopping (F = 1.776, p = .886). As a result, the 6 subjects were left in the sample investigated (for SPSS output file see Appendix D, table D4).

Considering the eliminations made from the original data set in accounting for privacy concerns and perception of the search engine, 150 observations remained for further analysis: 119 camera shoppers and 31 jacket shoppers.

5.1.2 Identical Store Design

As was outlined in section 3.3.3, it was vital to design the two store versions for cameras and winter jackets as similarly as possible so that navigational behavior can be attributed to the product nature and not to the store environment. As a result, navigational opportunities and product display were provided in the two store versions including a similar quantity of products on offer, a similar number of attributes used to describe each product and an identical breadth of agent communication.<32> All products had the same price range between 200 - 500 DM
($ 100 - 250).


First, it was ensured that satisfaction with agent communication would be comparable for the two store versions. This implied an emphasis on similar levels of performance of the search algorithms used in the two stores. For those subjects whose behavior has been considered in the analysis, there was no significant difference in satisfaction with agent Luci (z = -.353, p = .724) (for SPSS output file see Appendix D, table D5).

Second, the nature of information exchanged with the agent needed to be perceived similarly. Naturally, however, the nature of information exchanged between the agent and customers had to differ for compact cameras and winter jackets. An effort has therefore been made to align the perception of the communication process by ensuring that question legitimacy and importance would be distributed equally in the two store versions. For this purpose an independent pre-study was conducted where 39 subjects rated each one of the 56 agent questions (112 for both store versions) on a 10-point scale as to their perceived legitimacy and importance in an Internet sales context [Annacker et al., 2001]. Mann-Whitney U-test on the mean perceived question legitimacy of the 56 agent questions confirmed non-significant differences for the two store versions (z = -.867, p = .386). A T-test on mean perceived question importance of the two agent-question catalogues rendered a similar result (F = .577, p = .450). Thus, all in all, it seems that the degree of relevance and legitimacy inherent in the sales dialogue was perceived similarly for the two store versions (for SPSS output file see Appendix D, table D6).

Finally, the order in which agent questions would be asked was important, as it has been shown to influence navigation [Hoque and Lohse, 1997]. For this reason, communication was arranged identically in both store versions. It included 7 question cycles for each product with agent questions being arranged in each cycle in an order of decreasing importance.<33>


5.2 Choice and Perception of Products

To confirm the assumption that compact cameras and winter jackets would be perceived as search and experience goods respectively and entail different levels of uncertainty, several measures have been proposed by Weiber et al. [1995a]. Weiber et al. [1995a] argue that the degree to which a good can be considered a search, experience or trust good is founded on the uncertainty that a consumer perceives in judging the respective good‘s quality prior to purchase. Subjects who had come to purchase a winter jacket or compact camera were therefore asked how comfortable they felt (q1) and how probable it would be (q2) to fully judge upon the quality of the product they sought with the help of the Internet. In addition, they were asked how uncertain they felt in general that the product would meet their expectations (q3). The answers, which were given on a 6-point scale, are summarized in table 6. They show that participants felt in average less certain in the judgment of jackets. This perception of uncertainty comes close to statistical significance, however, only for q2. Cross-checking this finding with a larger data-set (where an additional 119 answers to questions q1 to q3 were available) improved the level of significance.<34> It can therefore be argued that the perception of winter jackets as an experience good, with slightly higher levels of purchase uncertainty, is supported by the data, if only weakly. Compact cameras, in contrast, are perceived as a search good with slightly lower levels of purchase uncertainty.


Tabelle 6: Perception of Experimental Products as Search or Experience Goods:

Questions employed to test perceived product nature as an experience or search good

Mean Value
Winter Jackets

Mean Value Compact Cameras

Statistics I
(sample: 150)

Statistics II
(sample: 269)

Q1: How comfortable are you that, with the help of the Internet, you‘ll be able to fully judge on all quality characteristics important to you [in the winter jacket]?
(1= not at all comfortable (..2,3,4,5)
6= very comfortable)



z = .726

p = .668

z = 1.505

p = .022

Q2: Please indicate, how probable it is that in the context of an Internet purchase you‘ll be able to fully judge on all quality characteristics important to you [in the winter jacket]?
(1= not at all probable (...2,3,4,5)
6 = very probable)



z = 1.339

p = .055

z = 1.459

p = .028

Q3: Please indicate on a 6-point scale how uncertain you generally feel now, before the purchase of a new winter jacket/compact camera, that [the product] will fully meet your expectations!
(1 = very uncertain (...2,3,4,5)
6 = not at all uncertain)



z = .414

p = .995

z = .759

p = .613

In addition to the perception of compact cameras and winter jackets as respective search and experience goods with different levels of purchase uncertainty associated to them, the two products were also chosen with a view to different types of risk dimensions expected to be dominant in them. As was described in detail in section 4.3.1., risk was broken down into four dimensions including functional, financial, sociological and psychological risk. Risk was calculated by multiplying the


perceived degree of loss and probability of loss for all four dimensions of risk and then summing. On an index level, cameras were perceived to be functionally more risky than jackets. More socio-psychological risk components were associated with the purchase of jackets.

Table 7 summarizes the product risk perceptions actually measured for cameras and jackets. It shows that the two products chosen for the experiment do, in fact, raise different buyer concerns. While compact cameras have a relatively high functional and financial risk, jackets display higher risk levels in the socio and psychological area. However, in contrast to expectations, the overall level of perceived risk (OPR) measured prior to purchase among participants was similar for the two products.<35>

The reason why so similar levels of perceived risk have been observed may have to be attributed to the self-selection process of experimental participants: only those people may have registered for the experiment that are already relatively open to use direct marketing channels such as the Internet and may for this reason be generally less risk averse.

As a result of similar OPR for the two products, observed differences in behavior that are reported on in this chapter cannot be directly attributed to different levels of OPR, but must be more seen in the light of distinct levels of uncertainty to judge on product quality prior to purchase. To a certain extend, of course, risk and uncertainty are related constructs as both integrate a ’probability-notion‘ of a loss to occur. This is mirrored in the significant bivariate correlations between OPR and the levels of uncertainty measured with CORRQ1 = -.218 (significant at p < .01) and
CORRQ2 = .-198 (significant at p < .05) (for SPSS output file see Appendix D, table D7). However, uncertainty does not respect the magnitude and relevance of loss to a consumer.


Tabelle 7: Perceived Risk Structure of Experimental Products:

Median of OPR
observed across
150 subjects

Perceived Functional

Perceived Financial

Perceived Psychological

Perceived Sociological


Winter jackets






Compact Cameras






Statistics: T-test for paired samples

(*Wilcoxon Test)

T = 4.380

p = .000

T = 3.738

p = .000

T = -4.349

p = .000

Z = -4.938*

p= .000*

T = -.343

p = .732

5.3 Observed Interaction Behavior

The first step to analyze the information search activity for the two product categories was to look at the total time users spent in the online store as well as the time expanded for the three distinct phases of the shopping session (orientation, dialogue and detailed product inspection).<36> In addition, some quantitative measures were considered to describe the way in which camera and jacket shoppers differed in their product inspection behavior. Table 8 gives an overview of the findings.

Table 8 shows that jacket shoppers in total invested around 19% more time (t) into the shopping trip than camera shoppers did; in average an additional 4,7 minutes. Particularly interesting in this context is to what part of the shopping session this time was dedicated. Obviously, participants interested in the experience good jacket attributed considerably more importance to manual product inspection. In average they spent 30% more time here than camera shoppers did (ti). Analyzing this behavior in more detail, jacket shoppers seem to have invested this time in a significantly larger number of objects viewed (73 versus 40) and more than twice as many photographs enlarged. However, they only required a fraction of time on individual objects when compared to camera shoppers.<37> Thus, jacket shoppers seem


to have quickly ’sifted through‘ the offer as a whole spending relatively little time per product and judging stronger on visual perceptions than camera shoppers who viewed much less products, but in average invested about twice as much time in the inspection of each individual product. The significantly larger time investment by camera shoppers per product indicates that they must have read most of the fact sheets and marketing texts presented for each product.

Tabelle 8: Comparison of Breadth of Interaction for Cameras and Jackets :

Interaction Indicators

(Mean Investment in Product Identification)







Level of Significance


Time Investment Measures:

- mean time investment, total (t)

- mean time for orientation (to)

- mean time for communication (td)

- mean time for detailed product
inspection (ti)


24,5 min (109)

0,7 min (112)

12,1 min (115)

11,5 min (120)


29,2 min (30)

0,4 min (36)

13,8 min (31)

14,9 min (37)






Manual Product Inspection:

- n° of products inspected

- time per product

- n° of photo enlargements



0,25 min




0,14 min






(* T-test; **Mann-Whitney U-test)

Besides these time variables, the overall findings summarized in table 8 suggest that jacket shoppers, who felt slightly less certain in the judgment of the product, displayed significantly higher levels of overall activity in the search process. At the same time, they searched in a different manner than camera shoppers did, relying more heavily on the manually controlled forms of search.


In order to better understand the type of interaction sought by the two shopping groups, two indices were developed. The first index, a communication quota (Qf), is a set-based measurement designed to express how much of the shopping process was generally dedicated to communicating with the agent versus obtaining information manually. A second index, a modification quota (MQR) was then used to analyze the dialogue that participants sought with the agent in more detail. The communication quota was defined as:


C =

total number of requests for a agent question page (including: those pages that were not answered and return hits to correct initial answers given, question category survey page and requests for Top-10 consideration set)

I =

total number of requests for pages giving product information, photo enlargements and required return hits to the top-ten set from both phases (orientation and product inspection)

As can be seen from table 9, camera shoppers have a significantly higher communication quota than jacket shoppers. This means that subjects searching for a camera relied relatively more on the exchange with the agent in their information search process than jacket shoppers did. Even though both groups of participants consulted the shopbot with a similar frequency (e.g. answered a similar amount of questions and made a similar number of modifications to initial specifications), jacket shoppers displayed a significantly higher need for manually controlled product inspection. Figure 6 visualizes these diverging navigational foci by giving a broad overview of the click streams that were observed for camera shoppers (above) and jacket shoppers (below) in the two versions of the online store.


Tabelle 9: Comparison of Depth of Interaction for Cameras and Jackets:

for Agent Interaction



Level of Sig.

- mean communication quota [Cf]

- share of questions answered

- median of modifications made [M]

- modification quotas for risk dimensions [MQR]





- modification quotas for privacy dimensions [MQP]
















































Abbildung 7: Users‘ Path through the Experimental Online Store <38>

As was outlined above, agent questions were not only product related, but also addressed the user in person and asked for the goals of search (e.g. desired use for the product). Given the wide spectrum of 56 agent questions, one goal of the current analysis was to find out what type of question people would be willing to answer while shopping for one or the other product. A correlation was therefore expected to be seen between the dominant risk dimensions of a product (e.g. social risk for jackets) and users‘ motivation to answer agent questions best suited to address them. However, as the 150 participants answered in average more than 85% of total agent questions, there would have been a strong ceiling effect present in the analysis of the number and share of questions answered. As a result, an attempt was made to ’grasp‘ users‘ qualitative purchase concerns in more detail by investigating the type of


question modified. For this purpose, a modification quota was developed for those subjects that did make adjustments to initial specifications to agent questions. As was described in section 3.3.4, agent questions referred to different risk dimensions and privacy classes that were used in the current analysis to determine a modification quota per question category (see also Appendix B3):



Average modification quota to be found in a question category R, where R refers to a bundle of questions addressing either functional (fun), financial (fin), social (soc) or psychological (psy) risk or where R refers to a bundle of questions that represent different privacy classes such as non-private questions relating directly to the product (pd), marginally private question indirectly referring to the product (pepr), purely personal questions (peip) or finally relatively private questions concerned with product usage (u)


Number of modifications made in one question category R by an individual i searching for a product p


Number of questions encountered by an individual i in a category R for a product p.


Number of individuals who shopped for product p and made modifications to any of the categories

The median of modifications made per product category (M) (see table 9) shows that jacket shoppers who modified agent options did so only slightly more often than their camera counterparts although this finding is not significant. This finding corresponds


to the fact that jacket shoppers also perceived slightly higher levels of uncertainty connected to their purchase.

Looking in more detail into the type of modifications made it turned out that in line with cameras‘ higher levels of inherent financial risk, shoppers for this product category also adjusted more often the agent‘s price parameters available in the search engine. More precisely, the data revealed that about 11% of camera shoppers adjusted the price range in which they wished to buy at least once while subjects searching for a jacket had in general a relatively firmer idea of what they wanted to spend (only 5% changed the price range once at a maximum).

Another finding that suggests perceived purchase risk to be in line with risk reduction behavior is the construct of psychological risk. Jacket shoppers modified significantly more agent questions that addressed this risk construct which was particularly relevant for jackets. Surprisingly, however, this type of consistent behavior could not be observed for the sociological risk dimension. Obviously, camera buyers did feel a need to modify just as many agent questions concerning ’social acceptance‘ of their product than jacket shoppers did (which is not in line with the level of sociological risk measured in advance of the shopping sessions).

Besides this comparison of perceived risk dimensions inherent in a product with subsequent attempts to address them during the information search process, it was also important to see what type of agent questions users would find important for product selection. Here the data suggest that consistent with the experience characteristic of apparels, jacket shoppers made significantly more modifications to usage related agent questions than camera shoppers did. In general, looking at the relative number of modifications made to personal and usage related questions, jacket shoppers seem to have put more emphasis on these relatively private issues of purchase than camera shoppers. Jacket shoppers were also significantly more willing to respond to private issues in the purchase context (“peip-questions“). Seen that usage related and personal questions were rated as rather illegitimate and unimportant in the independent study conducted (see section for analysis and Appendix C, table C1 for data), the modification quotas could suggest that users


allow for more insights into their private lives when product nature justifies this. More research is, of course, needed to confirm this preliminary evidence.

5.4 Discussion of Results for Online Marketing

The measured perception of products confirmed that participants felt slightly more uncertain in judging the quality of winter jackets prior to purchase. Jackets for the purpose of this study can therefore be regarded as a representative experience good. However, against expectations, the level of overall perceived risk was not significantly higher for jackets than for cameras. Therefore, the observed superior levels of interaction for winter jackets can not be attributed to the absolute amount of perceived risk prior to purchase (OPR). Instead, they seem to be more attributable to the ’experience‘ nature of the product, and the concurrent need of users to extensively inspect and visualize all product alternatives on offer (i.e., trying to anticipate the experience).

Clear support was rendered by the findings for the argument that consumers have distinct navigational needs when they search for different products online. In fact, today‘s electronic commerce environments display a strong lack of product context recognition. Not only do they often fail to support users effectively in their decision making process [Spiekermann and Parachiv, 2001], but site design and interactive functionalities also tend to follow an approach of ’one-size-fits all‘ for most product categories: Information provision is not always adjusted to those product attributes and features that might be of particular concern to customers. Usually, the same type of information is displayed no matter which product the online customer came for. Dialogue-systems strongly focus on product attributes only, but in general do not correspond to consumers‘ softer purchase concerns. Finally, detailed product representation, product description or visualization, are mostly identical in a domain for all goods on offer. The findings presented in this chapter show the necessity for online marketers to respect product nature more explicitly in the design of web sites. More specifically, the results include some hints for the design of agent dialogue design as well as context adjusted representation of products.


5.4.1 Product Related Focus of Dialogue Systems

It was shown that customers associate different types of purchase risk with the products they seek. In the current study, cameras were associated with a relatively higher functional and financial risk while jacket shoppers felt the socio-psychological side of the product to be relatively more important. In line with these product perceptions, camera shoppers also modified relatively more functional and financially related preferences. Considering the modification quota for questions with a socio-psychological focus jacket shoppers, in contrast, put significantly more focus on the appropriateness of these variables. As a result, some evidence is given to the argument that dialogue systems could be enhanced if they respected the risk dimensions inherent in a product [Spiekermann and Parachiv, 2001].

At the same time, it was interesting to see that jacket shoppers also put a relatively strong weight onto the modification of functional product attributes. This is surprising given that the relatively small stated risk perception on this dimension prior to purchase. However, given this finding, marketers offering a differentiated dialogue along risk dimensions might also be able to observe the ’true concerns‘ of their customers in this way. Seeing that users put weight on the specification of specific product attributes corresponding to particular types of product risk, marketers could learn about the true drivers of the purchase decision-making process and adjust risk-reducing dialogue-systems accordingly.

Finally, the results suggest that dialogue-systems can be relatively detailed and lengthy. Not only did online users specify many product attributes when they were involved in a high-involvement purchase (see the surprisingly big share of agent questions answered and additional modification rate), but they even displayed a readiness to adjust softer and more personal variables addressed by the agent. Even though the time manipulation of the experimental set-up might have led participants to browse and answer more questions than they would usually correspond to in ’real-world‘ online stores, this finding is important for two reasons: Firstly, the high level of disclosure suggests that people do not value their privacy as much as current household surveys often suggest. Second, lengthy dialogues do not seem to lead to customer annoyance or a loss of trust. In contrast, 77% of the users expressed


satisfaction or even high satisfaction with the search engine and many underlined their positive experience by written remarks in the debriefing questionnaire stating that they had perceived the system to be extremely “user friendly“, that they had felt “personally addressed“ and “well guided“. This is surprising, as that the agent dialogue involved an extremely exaggerated detail of product specification including many highly personal questions. All in all, the results suggest that there is a lot of room in dialogue-systems to exchange information with consumers without inducing a feeling of privacy intrusion among them.

5.4.2 Context Adjusted Representation of Products

During the observed shopping sessions, jacket shoppers displayed a significantly higher interest in the detailed and manually controlled inspection of products than camera shoppers did. They wished to view many more products and had a stronger need for visualization (photo enlargements). At the same time, the inspection of fact sheets seem to have had less importance for this group of buyers. In contrast, camera shoppers viewed much less products, but attributed a lot more attention to detailed information on each object (time per product). The results suggest that online consumers appreciate a differentiated way in which products are presented: while for some products, for which appearance is important, the investment might be worthwhile to present them with a strong visual focus employing interface technology that allows to view, enlarge and turn the product, these interface capabilities might not be necessary for buyers of search goods. In contrast, search goods that can be well described on the basis of plain product attributes and factual criteria may be better represented if the web site allowed for an objective inspection of product details in the form of fact sheets and comparison matrices. More research would be needed to confirm this finding.

5.5 Conclusion

All in all, the comparative analysis of search behavior for winter jackets and compact cameras suggests that higher levels of uncertainty in product judgement lead to more manual search. At the same time, relative importance of the agent is reduced. This finding is roughly in line with what was expected on the basis of equation model


results presented in chapter 4 and it suggests that agents are not equally important for all electronic commerce purchase environments. However, search behavior was only investigated for two products, compact cameras and winter jackets. More research would be needed if the current findings were to be generalized.

Finally, the hypothesis derived from the structural equation model that OPR leads to more product inspection versus the use of an agent could not be confirmed on the basis of the current data set, as different levels of OPR were not able to be measured for the two products under study. More research would therefore be needed here as well. Doing so, particular emphasis would have to be put on the selection process of experimental participants in order to avoid the same self-selection problems that were encountered in the experiment.



Motivation was measured on a 9-point scale with 1 = yes, would have very much liked to do something else instead of participating in the shopping experiment and 9 = no, would not at all have liked to do something else.


There were, however, 8 facts listed to describe major attributes of cameras and 6 to describe jackets.


Importance rating were taken from the independent pre-study (see Appendix C, table C1)


In table 2, 206 observations have been reported on that were collected in 4 treatments. As was mentioned there, two additional treatments were included in the experiment the results of which are not reported in this thesis. As these subjects, however, filled out the same questionnaires as the sample reported on this section their judgement of products can be included in the present analysis.


These general relationships which are measured here across the whole of 150 participants also hold true when product judgement of only those is considered who were going to purchase or shop for a respective product.


The respect of time required the same outlier analysis described in chapter 4.


It must be recognized here that jacket shoppers had only 6 product attributes displayed while camera shoppers had 8 of them. This means that the different times recorded for cameras and jackets could, strictly speaking, be attributed to this differing number of purchase arguments displayed. However, if the time per product is divided by the number of attributes viewed than there is still a significant difference between the time per product with jacket shoppers spending much less time per product (0,25 min per camera model/8 camera attributes = 0,031 min/attribute and 0,14 min per jacket model/6 jacket attributes = 0,025 min/attribute).


The figure presents ’stratograms‘ [Berendt, 2001] that trace users‘ paths through the site. The x-axis contains the steps in the navigation history, while the y-axis represents the type of page requested. Values along the y axis are ordered to reflect the interaction process: 0 is the question category survey page from where users can enter different cycles of agent questions, 1 to 4 is any question page, -1 is the display of product rankings, -2 is the detailed product description and -3 the respective photo enlargement. Navigation presented here starts with the communication phase2.


is divided through the number of questions in a category (Qi) in order to take account of the fact that the different question types (pd, pepr, peip, u or fin, func, psy, soc) were not distributed equally in the two store versions.

© Die inhaltliche Zusammenstellung und Aufmachung dieser Publikation sowie die elektronische Verarbeitung sind urheberrechtlich geschützt. Jede Verwertung, die nicht ausdrücklich vom Urheberrechtsgesetz zugelassen ist, bedarf der vorherigen Zustimmung. Das gilt insbesondere für die Vervielfältigung, die Bearbeitung und Einspeicherung und Verarbeitung in elektronische Systeme.

DiML DTD Version 2.0
Zertifizierter Dokumentenserver
der Humboldt-Universität zu Berlin
HTML - Version erstellt am:
Mon Dec 23 13:23:15 2002